Building ffmpeg with nvenc inside docker container

General
,
1

Hello! I’m making Dokerfile image for my python script + ffmpeg
I have docker image:

# Use the NVIDIA CUDA base image with the desired CUDA version
FROM nvidia/cuda:12.1.0-cudnn8-devel-ubuntu20.04

# Set environment variables to prevent interactive prompts during installation
ENV DEBIAN_FRONTEND=noninteractive

# Install dependencies and set timezone non-interactively
RUN apt-get update && apt-get install -y \
    tzdata \
    build-essential \
    pkg-config \
    git \
    wget \
    yasm \
    nasm \
    libx264-dev \
    libx265-dev \
    libnuma-dev \
    libvpx-dev \
    libfdk-aac-dev \
    libmp3lame-dev \
    libopus-dev \
    libass-dev \
    libfreetype6-dev \
    libfontconfig1-dev \
    python3.9 \
    python3.9-venv \
    python3.9-dev \
    curl \
    htop \
    && ln -fs /usr/share/zoneinfo/America/New_York /etc/localtime \
    && dpkg-reconfigure --frontend noninteractive tzdata \
    && rm -rf /var/lib/apt/lists/*

# Update alternatives to set python3.9 as default
RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1 \
    && update-alternatives --set python3 /usr/bin/python3.9

# Install pip for Python 3.9
RUN curl https://bootstrap.pypa.io/get-pip.py | python3.9

# Set environment variables for CUDA
ENV PATH=/usr/local/cuda/bin:${PATH}
ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:${LD_LIBRARY_PATH}

# Install Jupyter and other Python dependencies
RUN pip3 install jupyter jupyterlab

# Clone FFmpeg source
RUN git clone https://github.com/FFmpeg/FFmpeg.git /ffmpeg
RUN git clone https://github.com/FFmpeg/nv-codec-headers.git /nv-codec-headers

# Install nv-codec-headers
WORKDIR /nv-codec-headers
RUN make
RUN make install

# Build FFmpeg with NVIDIA support
WORKDIR /ffmpeg
RUN ./configure \
    --prefix=/usr/local \
    --enable-cuda-nvcc \
    --enable-cuvid \
    --enable-nvenc \
    --enable-nonfree \
    --enable-libnpp \
    --extra-cflags=-I/usr/local/cuda/include \
    --extra-ldflags=-L/usr/local/cuda/lib64 \
    --enable-gpl \
    --enable-libx264 \
    --enable-libx265 \
    --enable-libvpx \
    --enable-libfdk-aac \
    --enable-libmp3lame \
    --enable-libopus \
    --enable-libass \
    --enable-libfreetype \
    --enable-libfontconfig \
    && make -j$(nproc) \
    && make install \
    && make clean

# Install TensorRT dependencies
RUN apt-get update && apt-get install -y \
    libnvinfer8 \
    libnvonnxparsers8 \
    libnvparsers8 \
    libnvinfer-plugin8 \
    && rm -rf /var/lib/apt/lists/*

# Install TensorRT Python package
RUN pip3 install nvidia-pyindex \
    && pip3 install onnxruntime-gpu \
    && pip3 install opencv-python \
    && pip3 install onnx

# Verify installation
RUN ffmpeg -version

# Expose the Jupyter Notebook port
EXPOSE 8888

# Set the working directory
WORKDIR /superres

# Set entrypoint to Jupyter lab
ENTRYPOINT ["jupyter", "lab", "--ip=0.0.0.0", "--port=8888", "--no-browser", "--allow-root", "--NotebookApp.token=''", "--NotebookApp.iopub_data_rate_limit=1.0e10"]

And I have few problems:
1)When I run this command: ffmpeg -i test_video_720p.mp4 -c:v h264_nvenc -preset slow -b:v 5M output.mp4 I get this:

image
image2053×394 110 KB

And I cant find the solution.
Also I have a problem - I can’t upload files more then 1 MB on server + my terminal works strange (I cant use arrows - I get ^[[A)

2

It seems that your command requires an nvidia GPU and as the error messsage shows, libraries are missing too. Looks like somebody already built an image for ffmpeg and nvenc. You can check the description for ideas about using a gpu in a container

https://hub.docker.com/r/willprice/nvidia-ffmpeg

Note: Don’t share screenshots of texts. Harder to see and parts of it cannot be quoted or copied to search for.

Regarding the upload limit, that must be limited by the application / proxy server you are using. Docker will not limit it.

1 Like